EP0375352B1 - Procédé pour chercher une matrice de données binaires - Google Patents

Procédé pour chercher une matrice de données binaires Download PDF

Info

Publication number
EP0375352B1
EP0375352B1 EP89313265A EP89313265A EP0375352B1 EP 0375352 B1 EP0375352 B1 EP 0375352B1 EP 89313265 A EP89313265 A EP 89313265A EP 89313265 A EP89313265 A EP 89313265A EP 0375352 B1 EP0375352 B1 EP 0375352B1
Authority
EP
European Patent Office
Prior art keywords
scans
search
scan
matrix
binary
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
EP89313265A
Other languages
German (de)
English (en)
Other versions
EP0375352A1 (fr
Inventor
Stewart Beaton Kelland
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NCR International Inc
Original Assignee
NCR International Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NCR International Inc filed Critical NCR International Inc
Publication of EP0375352A1 publication Critical patent/EP0375352A1/fr
Application granted granted Critical
Publication of EP0375352B1 publication Critical patent/EP0375352B1/fr
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition

Definitions

  • This invention relates to a method of searching a matrix of binary data.
  • the image data consists of a plurality of pixels or picture elements obtained at a scanning line associated with the scanning device.
  • the pixels associated with successive scans obtained at the scanning line comprise the matrix of image data associated with the item or document.
  • the image data is processed by conventional circuitry to eliminate "noise” and the like, and to also threshold the image data to produce a binary matrix of data.
  • each pixel of image data which may comprise an eight bit gray scale value which is obtained from the scanning line, is reduced, by thresholding, to a binary "1" or a binary "0".
  • a binary 1 may indicate the presence of data, while a binary 0 indicates the background or the absence of data.
  • the image data about a document is thresholded to a matrix of binary data.
  • One of the problems with using a matrix of binary data is that it is often difficult to find what is looked for within the matrix. In other words, there may be a lot of extraneous binary data in the matrix in addition to what is sought. For example, one may look for certain lines in the matrix of data. In this regard, long lines which are parallel to the original direction of scanning are not difficult to find, whereas, long lines which are perpendicular to the original scan line are difficult to find. For example, if the scanning is done along successive columns, long lines which are parallel to the direction of scanning are easy to find. In other words, a long string of binary ones will be found in a particular column or adjacent columns, indicating the presence of a long line. However, if the long line is perpendicular to the original scanning direction, it means that a portion of the line will extend over many columns of data, making the search for this long line difficult.
  • a method of searching a matrix of binary data derived by scanning a document along successive scan lines which are parallel to a given direction characterized by the steps of: determining a number of linear search scans parallel to said given direction to cover a predetermined area of said matrix of binary data and searching for the presence of binary ones along said predetermined number of search scans to generate search results for each of said search scans, with a binary one indicating the presence of data; subjecting the search results of preselected said search scans to an AND logic operation to generate AND-scans; subjecting preselected said AND-scans to an OR logic operation to generate OR-scans; and using said OR-scans to define search regions in which to look for straight lines which are perpendicular to said given direction.
  • Another advantage of the present invention is that it facilitates the identification of certain documents. For example, certain lines appearing on certain locations of the document may be indicative of that document being a particular type of business form. The lines may be part of a general quadrilateral box or area, for example, located on the document. These long lines may be parts of "boxes" on a form, "underlines”, or "separators".
  • Finding the long lines facilitates optical character recognition (OCR) in that the long lines are generally removed first before character recognition is attempted.
  • OCR optical character recognition
  • the present invention relates to a process for searching a matrix of binary pixel data associated with a document so as to find search regions in which straight lines may be found, with the straight lines being those which are perpendicular to the original direction of scanning which produced the matrix of binary data. Once the search regions are found, they can be used to focus the searches within the search areas to locate the long lines.
  • Fig. 1 shows apparatus 10 which may be used in carrying out the invention.
  • the apparatus 10 includes an item transport 12 which moves an item, like a document 14, towards a scanning line 16 where the document 14 is imaged by a scanner 18 as is conventionally done.
  • the scanner 18 produces successive scan lines or columns of pixel data or pixels as the item 14 is moved in reading relationship therewith.
  • the scanner 18 could also be a hand held scanner, for example, which is moved over a stationary document to effect the reading.
  • From the scanner 18, the successive columns of pixel data are processed to minimize noise associated with the pixels and to threshold the pixels into a binary "1", for example, which may represent the presence of data and a binary "0" which may represent the absence of data.
  • This processing is effected by conventional circuitry shown only as processing circuitry 20.
  • the output of the processing circuitry 20 is a matrix 22 of binary data or pixels (not compressed) which corresponds to the image of the associated document 14.
  • the matrix 22 of binary data may include about 1700 pixels per column with a resolution of 200 pixels per inch (25.4mm), for example, as just one illustration of a document whose width is about 8 1/2 inches (216mm) wide.
  • the scanner 18 and the item transport 12 may be controlled by a separate controller 24, for example, or they may be controlled by the controller 26 which is used to process the matrix 22 of data or pixels.
  • the controller 26 (Fig. 1) is a conventional controller which may be used to process the matrix 22 of pixels according to this invention.
  • the controller 26 includes a read only memory (ROM 28), a random access memory (RAM 30), a key board (KB) 32, a display 34, interfaces 36 and 38, and interface and control logic 40 which is used to interconnect all the components shown in a conventional manner.
  • ROM 28 read only memory
  • RAM 30 random access memory
  • KB key board
  • the form of the controller 26 shown is used to simply facilitate a discussion of the operation of the controller 26;
  • column 0 may start at the top right side of the document 14, as viewed in Fig. 1, while column 1,000 approaches the left side of the document 14.
  • row 0 appears at the top of the document 14, while row 1700 approaches the bottom of the document 14, as viewed in Fig. 1.
  • the upper left hand corner of the matrix 41 corresponds to the upper right hand corner of the document 14 shown in Fig. 1.
  • the columns and rows may be reversed, depending upon how the documents are scanned, for example.
  • the lines and printing shown in Fig. 2 represent the binary ones of data which comprise the matrix 41, with the binary zeros comprising the background of the document 14. It is the lines like lines 42, 44, 46, 48, and 50, for example, which are difficult to find; these are the lines which are perpendicular to the direction of scanning line 16.
  • the first step in the process or method according to this invention is to determine a number of search scans which will be used to search the matrix 41 of binary data. These search scans will be used to search the matrix 41 in a direction which is parallel to the scanning line 16 which is also shown on Fig. 2. The binary ones found in these search scans, after some processing to be later described herein, may be indicative of the long lines being sought.
  • the next step according to this invention is to scan a predetermined area of the matrix 41 of binary data in a direction which is parallel to the original direction of scanning (scanning line 16).
  • the predetermined area may be a portion of the matrix 41, or it may be the entire area of the matrix 41.
  • the predetermined area relates to the entire matrix 41.
  • N was equal to five.
  • the search results for the search scans #1 through #5 shown in Fig. 2 are shown as scans-1 through 5, respectively, in Fig. 3.
  • binary data 42-1 shown in Fig. 3 for scan-1 corresponds to a portion of the line 42 shown in Fig. 2.
  • binary data 44-1 corresponds to a portion of line 44.
  • lines 46 and 50 are thicker than the lines 42 and 44; correspondingly, the associated binary data 46-1 and 50-1 is "wider", as measured along the search scan #1, than the data 42-1 and 44-1 associated with lines 42 and 44-1.
  • the next step in the method according to this invention is to AND certain of the scans-1 through -5 shown in Fig. 3 in the following manner.
  • the scans-1 and -2 which comprise an AND-set are ANDed together to generate the AND-scan-1 shown on line 56 in Fig. 3.
  • the digital data 42-1-2 is the result of ANDing digital data 42-1 and 42-2.
  • Scans-2 and -3 are similarly ANDed together to generate AND-scan-2.
  • the digital data 42-2-3 is the result of ANDing digital data 42-2 and 42-3.
  • Scans-3 and -4 are ANDed together to generate AND-scan-3
  • scans-4 and -5 are ANDed together to generate AND-scan-4.
  • the next step in the method of this invention is to "OR" the AND-scan-1 through AND-scan-4 data in the embodiment described.
  • the digital data (actually binary 1s) associated with the AND-scan-1 and AND-scan-2 are ORed together to generate the "OR-scan-1" shown in Fig. 4.
  • a function of the OR operation is to catch a long line which may be discontinuous at certain points along its length.
  • the AND-scan-1 and the AND-scan-2 comprise an "OR set” alluded to earlier herein.
  • AND-scan-2 and AND-scan 3 are ORed together to generate OR-scan-2, and correspondingly, AND-scan-3 and AND-scan-4 are ORed together to generate OR-scan-3.
  • This process is repeated for additional AND-scans which may appear in a particular application to be processed. Again, there is some overlapping in the method in that the AND-scan-2 is used in the generation of OR-scan-1 and OR-scan-2.
  • the OR-scans-1 through -3 are used to define search regions in which the controller 26 may look to find the long lines, like lines 42 and 44 for example, which are perpendicular to the direction of scanning line 16.
  • the controller 26 keeps a record of the row and column positions for the binary data derived from the search scans #1 through #5 as discussed in relation to Figs. 2-4.
  • the digital data 42-OR-1 (Fig. 4) is derived from the AND-scan-1 and the AND-scan-2, which in turn, are derived from scan-1, scan-2 and scan-3 as previously described.
  • the controller 26 then extends the search region associated with OR-scan-1 by extending it to cover a distance including search scans #1-#3 and a little distance on each side of search scan #1 and search scan #3 as shown by bracket 58 shown in Fig. 5.
  • OR-scan-2 extends at least over a distance including search scans #2-#4 as indicated by bracket 60.
  • OR-scan-3 extends at least over the distance indicated by dashed bracket 62.
  • the location of the binary data 42-1 from scan-1 and the location of binary data 42-5 from scan-5 shown in Fig. 4 are used to establish the row limits for the search region 42-R shown in Fig. 5. Search region 42-R (Fig.
  • search region 5 is made slightly wider than that warranted by the width of the data, like 42-1 (Fig. 3), itself, so as to provide a rectangular search region as shown to make sure that line 42 shown in Fig. 2 is included in the search region 42-R.
  • the remaining search regions 44-R, 46-R, 48-R and 50-R shown in Fig. 5 are similarly determined, and they correspond to the long lines 44, 46, 48, and 50, respectively, shown in Fig. 2.
  • the search regions, like 42-R are then used to locate the long lines which are perpendicular to the scanning direction or scanning line 16.
  • a search region for the wide line 54 shown in Fig. 2 is not included in Fig. 5.
  • the digital data, like 54-OR-3 shown in Fig. 4, is too wide to be indicative of a narrow line; consequently, it is discarded by the controller 26 instead of being used by it to create a search region as described.
  • squiggle or illegible scrawl 64 shown in Fig. 2 and shown as 64-1 in Fig. 3 is eliminated in AND-scan-3 shown in Fig. 3.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Character Input (AREA)
  • Image Analysis (AREA)

Claims (5)

  1. Une méthode de recherche d'une matrice de données en binaire dérivée en balayant un document (14) le long de lignes de balayage successives (16) qui sont parallèles à un sens donné, caractérisée par les étapes: de détermination d'un nombre de balayages de recherche linéaires parallèles audit sens donné pour couvrir une zone prédéterminée (41) de ladite matrice de données en binaire et de recherche de la présence de "1" binaires le long dudit nombre prédéterminé de balayages de recherche pour générer des résultats de recherche pour chacun desdits balayages de recherche, un "1" binaire indiquant la présence de données; de soumission des résultats de recherche desdits balayages de recherche présélectionnés à une opération logique ET pour générer des balayages ET; de soumission desdits balayages ET présélectionnés à une opération logique OU pour générer des balayages OU; et d'utilisation desdits balayages OU pour définir des régions de recherche (42-R à 50-R) dans lesquelles on recherchera des lignes droites (42-50) qui sont perpendiculaires audit sens donné.
  2. Une méthode conformément à la revendication 1, caractérisée en ce que ladite étape de détermination est effectuée en considérant la longueur des lignes droites (42-50) à trouver.
  3. Une méthode conformément à la revendication 2, caractérisée en ce que ladite étape d'opération logique ET est effectuée en soumettant au moins deux desdits balayages de recherche adjacents à une opération logique ET.
  4. Une méthode conformément à la revendication 3, caractérisée en ce que ladite étape d'opération logique OU est effectuée en soumettant au moins deux desdits balayages ET adjacents à une opération logique OU.
  5. Une méthode conformément à la revendication 1, caractérisée en ce que ladite étape d'utilisation est efficace pour définir les régions de recherche généralement rectangulaires (42-R à 50-R).
EP89313265A 1988-12-19 1989-12-19 Procédé pour chercher une matrice de données binaires Expired - Lifetime EP0375352B1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US286411 1981-07-24
US07/286,411 US4901365A (en) 1988-12-19 1988-12-19 Method of searching binary images to find search regions in which straight lines may be found

Publications (2)

Publication Number Publication Date
EP0375352A1 EP0375352A1 (fr) 1990-06-27
EP0375352B1 true EP0375352B1 (fr) 1994-03-02

Family

ID=23098486

Family Applications (1)

Application Number Title Priority Date Filing Date
EP89313265A Expired - Lifetime EP0375352B1 (fr) 1988-12-19 1989-12-19 Procédé pour chercher une matrice de données binaires

Country Status (5)

Country Link
US (1) US4901365A (fr)
EP (1) EP0375352B1 (fr)
JP (1) JPH02224187A (fr)
CA (1) CA1316605C (fr)
DE (1) DE68913475T2 (fr)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2683008B2 (ja) * 1988-02-16 1997-11-26 株式会社リコー 矩形図形内部塗りつぶし方法
JP2812982B2 (ja) * 1989-04-05 1998-10-22 株式会社リコー 表認識方法
JP2940936B2 (ja) * 1989-06-06 1999-08-25 株式会社リコー 表領域識別方法
JP3176052B2 (ja) * 1990-04-27 2001-06-11 キヤノン株式会社 画像処理装置
US5048099A (en) * 1990-05-21 1991-09-10 Eastman Kodak Company Polygon-based method for automatic extraction of selected text in a digitized document
EP0584559A3 (en) * 1992-08-21 1994-06-22 United Parcel Service Inc Method and apparatus for finding areas of interest in images
US5268580A (en) * 1992-09-02 1993-12-07 Ncr Corporation Bar code enhancement system and method for vision scanners
US5444793A (en) * 1993-06-15 1995-08-22 Ncr Corporation Method for detecting machine printed monetary amounts in binary images
US8000535B2 (en) * 2007-06-18 2011-08-16 Sharp Laboratories Of America, Inc. Methods and systems for refining text segmentation results
CN102156884B (zh) * 2011-04-25 2012-12-12 中国科学院自动化研究所 一种直线段检测和提取方法
CN104751177B (zh) * 2015-03-26 2018-01-12 常州大学 一种用于标识数字图像中给定模糊数字直线段的方法
CN105513044B (zh) * 2015-11-20 2018-07-17 常州大学 一种基于统计度量直线特征的数字直线段识别方法

Family Cites Families (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4044327A (en) * 1973-08-01 1977-08-23 Bunker Ramo Corporation Trimming potentiometers with coarse and fine adjustment means
US3930237A (en) * 1974-03-07 1975-12-30 Computervision Corp Method for automating the production of engineering documentation utilizing an integrated digital data base representation of the documentation
JPS57150075A (en) * 1981-03-12 1982-09-16 Fuji Xerox Co Ltd Square figure recognizing device
JPS58103266A (ja) * 1981-12-15 1983-06-20 Toshiba Corp 文字画像処理装置
US4566126A (en) * 1982-04-30 1986-01-21 Fuji Electric Company, Ltd. Pattern discriminator
US4516266A (en) * 1982-12-17 1985-05-07 International Business Machines Corporation Entity control for raster displays
JPS61203785A (ja) * 1985-03-07 1986-09-09 Dainippon Screen Mfg Co Ltd ２値画像デ−タの平滑化処理方法及びその装置
US4673987A (en) * 1985-04-22 1987-06-16 Ibm Corporation Compressed block code for facsimile transmission
JPH0750483B2 (ja) * 1985-05-22 1995-05-31 株式会社日立製作所 文書画像追加情報の蓄積方法
US4742399A (en) * 1985-11-01 1988-05-03 Dainippon Screen Mfg. Co., Ltd. Method of and apparatus for generating high-resolution bilevel image data
JPH0731714B2 (ja) * 1986-05-29 1995-04-10 インタ−ナショナル ビジネス マシ−ンズ コ−ポレ−ション 文字成分切出し方法
US4717962A (en) * 1986-12-17 1988-01-05 Kabushiki Kaisha Toshiba Method for compressing and reconstructing a data representation of a picture
US4728784A (en) * 1987-01-30 1988-03-01 Federal Express Corporation Apparatus and method of encoding and decoding barcodes

Also Published As

Publication number Publication date
EP0375352A1 (fr) 1990-06-27
US4901365A (en) 1990-02-13
JPH02224187A (ja) 1990-09-06
DE68913475D1 (de) 1994-04-07
DE68913475T2 (de) 1994-10-06
CA1316605C (fr) 1993-04-20

Similar Documents

Publication Publication Date Title
US5444793A (en) Method for detecting machine printed monetary amounts in binary images
US4932065A (en) Universal character segmentation scheme for multifont OCR images
JP2986383B2 (ja) ライン走査画像のためのスキューを補正する方法及びその装置
US6738154B1 (en) Locating the position and orientation of multiple objects with a smart platen
US4162482A (en) Pre-processing and feature extraction system for character recognition
EP0669593B1 (fr) Procédé de reconnaissance de code bidimensionnel
US5563403A (en) Method and apparatus for detection of a skew angle of a document image using a regression coefficient
US4817171A (en) Pattern recognition system
US4847912A (en) Method of detecting a space between words with optical character reader
US5915039A (en) Method and means for extracting fixed-pitch characters on noisy images with complex background prior to character recognition
US6438265B1 (en) Method of binarization in an optical character recognition system
EP0375352B1 (fr) Procédé pour chercher une matrice de données binaires
US5553162A (en) Method for detecting ink jet or dot matrix printing
US4242734A (en) Image corner detector using Haar coefficients
EP0975146B1 (fr) Détermination de la position et de l'orientation de plusieurs objets à l'aide d'un plateau d'entrée d'image intelligent
EP0702320A1 (fr) Détection d'inclinaison
JPS61289476A (ja) 文字読取装置のフオ−マツト生成方式
EP0974931A1 (fr) Procédé et dispositif pour l'identification de plusieurs sous-images dans une image d'entrée
JP3020293B2 (ja) 属性判別方法
JP2812705B2 (ja) 文字切出し装置
JPH0685186B2 (ja) 写真領域識別装置
JPH07111738B2 (ja) 文書中の領域境界抽出方式
JPH0334112B2 (fr)
JPS6343788B2 (fr)
JPS596419B2 (ja) 文字切出し方式

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): DE FR GB

17P Request for examination filed

Effective date: 19901106

17Q First examination report despatched

Effective date: 19921023

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: NCR INTERNATIONAL INC.

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): DE FR GB

REF Corresponds to:

Ref document number: 68913475

Country of ref document: DE

Date of ref document: 19940407

ET Fr: translation filed
PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed
REG Reference to a national code

Ref country code: GB

Ref legal event code: IF02

REG Reference to a national code

Ref country code: GB

Ref legal event code: 746

Effective date: 20030924

REG Reference to a national code

Ref country code: FR

Ref legal event code: D6

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: GB

Payment date: 20040923

Year of fee payment: 16

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: FR

Payment date: 20041029

Year of fee payment: 16

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: DE

Payment date: 20041110

Year of fee payment: 16

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: GB

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20051219

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: DE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20060701

GBPC Gb: european patent ceased through non-payment of renewal fee

Effective date: 20051219

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: FR

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20060831

REG Reference to a national code

Ref country code: FR

Ref legal event code: ST

Effective date: 20060831